Synthetic W-2 Wage and Tax Statement Data
Synthetic training data — no real PII, fully coherent identities
Generate synthetic 2025 four-up W-2 sheets — the current-year quarter-sheet layout with the employee's four copies on one page. The last four-up revision before the 2026 Box 14 split, and the counterpart to the 2024 sheet in any two-year income lookback.
152
Fields per document
1
Page
Tax
Category
What this document is
This is the 2025 revision of the four-up W-2 sheet: one letter-size page carrying four complete Wage and Tax Statements for the same employee — Copy B, Copy C, and two Copy 2 forms — perforated for the recipient to separate. It is the format most employees hold in their hands for the 2025 tax year, and the format most likely to be photographed and uploaded to a lender, a tax-prep product, or a benefits portal.
Why generate synthetically
Current-year documents are the ones a production pipeline sees most and the ones a training corpus is least likely to contain, because assembling a labelled corpus takes longer than a filing season. Generating the current revision closes that lag: you can evaluate against the 2025 sheet before a single real 2025 sheet has been collected, labelled and cleared for use. And because a W-2 concentrates a Social Security number, a name, an address and an income figure in one place — four times over on this layout — waiting for real documents means waiting for documents you will never be allowed to share anyway.
What makes synthetic data useful
Each sheet replicates one simulated employee's statement across four quadrants with identical values and distinct coordinates. Box 3 Social Security wages cap at the 2025 wage base of $176,100 while Box 5 Medicare wages continue past it, so the same high earner shows a different Box 3 on the 2025 sheet than on the 2024 one while everything else about them stays fixed — a controlled year-over-year contrast that is essentially impossible to construct from collected documents. Box 12 codes follow the realistic distribution of elective deferrals, health coverage cost, and Roth contributions rather than a uniform draw.
Training challenges
Everything hard about the four-up layout applies here: four instances of every label on one page, values identical across quadrants so cross-quadrant merges are invisible to value-only scoring, quarter-page boxes with correspondingly tiny checkbox targets, and quadrant boundaries marked by thin cut lines rather than whitespace. The 2025 revision adds a versioning problem on top. It is the last four-up before the 2026 Box 14a/14b split, so a pipeline in production during 2026 sees two structurally different Box 14 regions in the same inbox, and a model trained only on the current revision regresses on the one that arrived last week.
Generate synthetic W-2 Wage and Tax Statement data
Start with 500 free credits. No credit card required.
Generate NowWho uses this data
Mortgage and consumer-lending platforms verifying current-year income, tax-preparation software preparing for the filing season before real documents exist, payroll bureaus validating their own print output, and income-verification services whose accuracy claims are made against the current revision. Fraud teams use the year-over-year pair to train detectors on wage figures that move implausibly between revisions.
Document complexity profile
152 fields on a single page: 72 currency amounts, 60 text, 12 checkbox targets, and 4 Social Security number and 4 employer identification number typed fields, arranged as one 38-field statement replicated four times. 132 annotation relations, 16 conditional bindings, 20 function calls, maximum expression depth 2. Identical in structure to the 2024 sheet, which is what makes the pair a clean controlled comparison.
Key stats from our synthetic corpus
Quantitative characteristics of the W-2 Wage and Tax Statement documents our generator produces.
| Metric | Value | Detail |
|---|---|---|
| 2025 Social Security wage base | $176,100 | Box 3 Social Security wages top out at exactly $176,100 while Box 5 Medicare wages continue to $316,900 for the same high earners. The cap is the one value that reliably distinguishes the 2025 revision from the 2024 one on an otherwise identical layout. |
| Median Box 1 wages | $45,600 | Across 641 eligible synthetic identities the median Box 1 wage is $45,600, p25–p75 $27,200 to $74,400, spanning $4,000 to $316,900. Holding the wage distribution fixed across revisions is what makes the year-over-year comparison controlled. |
| Box 12 code distribution | D 72% / E 18% / G 10% | Box 12a codes are 71% D (401(k) elective deferrals), 17% E (403(b)) and 12% G (457(b)), populated on 57.1% of sheets. Box 12b is 83% DD (employer-sponsored health coverage cost) and 17% C, populated on 49.8%. Uniform sampling of codes teaches a model the wrong prior; this is the skew a natural corpus has. |
| Blank field rate | 42% | 64 of the annotated fields render blank on a typical sheet, repeated across all four copies. Every blank field is still annotated, so blank-versus-zero discrimination can be trained and scored rather than assumed. |
| Statements per sheet | 4 | Four complete statements with identical values and four distinct coordinate sets. Copy-level annotations let you measure whether an extractor returned one wage record or four — the deduplication error that value-only evaluation cannot detect. |
How this document co-occurs with others
Rates at which identities in our corpus that produce a W-2 Wage and Tax Statement also produce other documents.
| Correlation | Rate | Detail |
|---|---|---|
| Prior-year sheet, identical layout | 100% | The same employee's 2024 four-up sheet, differing only in the Social Security wage base. The pair isolates year-indexed validation logic from layout learning. |
| Following-year revision with the Box 14 split | 100% | The 2026 revision splits Box 14 into 14a and 14b for Treasury Tipped Occupation Codes. Pairing 2025 with 2026 is the structural-change test that a same-year corpus cannot provide. |
| Withholding election on file | 100% | The same employees have W-4s with their employer. Pairing the withholding election with the resulting Box 2 federal tax withheld is how onboarding and payroll systems validate that an election was actually applied. |
| Employer's annual unemployment return | 100% | The issuing employer also files Form 940. Aggregating an employer's W-2 wages against its FUTA return is the second half of the payroll reconciliation that starts with the 941. |
Wage and rate figures above are corpus-derived: they were computed over the 641 employment-eligible identities inside a local synthetic corpus of 1,000 identities generated by SymageDocs' World Simulation Engine at seed 20260421. Field, type and relation counts come from the shipped 2025 four-up W-2 definition in the SymageDocs form library. No real employee, employer, or payroll data was used at any stage.
Frequently asked questions
- What data format do synthetic 2025 four-up W-2 documents include?
- Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes, field types, and ground-truth values for all 152 fields on the sheet — 38 fields per copy across four copies — plus 132 label-to-value relations, each recording which copy it belongs to. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.
- What actually changed between the 2024 and 2025 four-up sheets?
- The layout is unchanged and the Social Security wage base moved from $168,600 to $176,100. That is a small change with a large consequence for validation logic: a rule that hardcodes the cap rejects valid 2025 documents, and a model that learned the cap as a value rather than as a year-indexed parameter will mis-normalise them. Generating both years for the same employee isolates exactly that failure.
- Should I train on this or on the 2026 revision?
- Both, and evaluate them separately. The 2026 revision splits Box 14 into 14a 'Other' and 14b for Treasury Tipped Occupation Codes. Any pipeline running in 2026 receives 2025 and 2026 documents simultaneously, so a single-revision training set guarantees a regression on whichever revision you left out.
- Are the wage figures internally consistent?
- Yes. Each sheet belongs to a simulated person with a job, an employer, a state of residence and a benefits profile, so Box 4 Social Security tax is the correct rate applied to Box 3, Box 6 Medicare tax follows Box 5, and Box 17 state tax is zero for residents of states with no income tax rather than randomly small.
- Can I use this data commercially?
- Yes. Every Social Security number, name, address, employer and wage figure is synthetic, contains no real personal data, and is licensed for commercial use including model training, benchmarking, and redistribution inside your own products.